Activity Feed › Discussion Forums › GNSS & Geodesy › ALTA Measurement Standards
This is an area I have no practical experience with yet. This thread is hitting close to home, as it is making me feel quite self conscious about my lack of knowledge in this area. Does anyone have a link or can offer some direction on this. I have access to TBC, but I am not proficient with it yet, but am wanting to learn. Should I use another program that may be more intuitive for beginners? I need a tutor in person, but I don’t know that I have access to that currently.
My goal is to be able to run the required checks to be certain we are meeting standards, even if just for my own knowledge. Someone here may be doing this and I am just not aware.
after a proper weighted least squares adjustment .
It should read “proper weighted least squares with sufficiently redundant data.” There really isn’t enough data if you have few measurements beyond what is needed for a bare solution. This will be in part reflected in larger error ellipses.
The redundancy needs to be spread around in the network. I can measure one line 10 times and improve the printed numbers, but that doesn’t give a realistic answer for the rest of the network. Cross ties are great for getting a realistic solution. Forward and backward distances are good. Repeating all angles is good.
.@bill93 so true. At minimum We have 4 rounds for robot and i try and teach the guys if they throw a fly and to take a little extra time and if they can set the fly where they can set it to see another previously measured control point or another corner etc then do it. I. Other words on top of repeating measurements on conventional traverse do cross ties or whatever. I have set random tripods up in middle of a field not needed for the boundary but it allows me to turn rounds or sets as I move around the property so i build more redundancy and more of a network. Rtk base and rover and robot used on same job can really allow so creative networks and time savings all in one. TBC and Trimble access truly make this easy. 90% of the time when we have an alta it is usually also for an engineering project so topo also. This makes it even nicer because we need control in more locations so i say no fly but you can get away with point only closures as you set additional control. The hardest thing to get through the field guys heads is they have this thing stuck in their heads that measuring rounds they have to only do bs and fs. Trying to teach them that you have your bs and fs set up set up on that corner and that corner and measure all rounds in one set up at same time. No need to separate them access can handle many additional points and once you have the first shot on them its all automatically done anyway. I did one last year around a park i had 4 rods and bi pods and 4 sets of legs total. Every corner on the lots where shot at minimum twice from different set ups as we traversed . I could set up 3 bi pods before the i man was ready and i did a lot of walking but the redundancy and such was awesome. It was a bit over kill but i used that as a teaching tool is all. It is a different way of thinking to build networks but if its done you almost never have to go back and you have confidence in all the measurements. I was taught years ago that that boundary survey measurements wise is only as good as your worst side shot. It doesn??t matter how good your loop closure was if you bust a angle to a corner . Probably not as common with today??s technology but happened years ago. Lol.
@jbw Trimble has some good power hours free to watch on network adjustments. They will go over the buttons and flow. Nothing better than reading some good books on least squares as that will get you the concept and terminology down and the understanding of the statistics you will see in reports. Thats the adjustment part. Then the alta report is simple in TBC. I run the alta report even when i am not doing an alta. It sometimes forces me to go back and check my adjustment in spots when I fail on certain point pairs.
At minimum We have 4 rounds for robot
I’m going to be that guy just for a moment here… 4 rounds is great and all, but in terms of a LSA it doesn’t do much besides reduce standard errors for ONE mean angle. Lots of folks say they have “sufficient redundancy” because they ran rounds. In my opinion: nope. At that rate you’re just applying centering errors to a traverse. That’s it.
I personally instruct our field crews to instead do 1-2 rounds, break setup, relevel with a different HI, and then do an additional 1-2 rounds. That way you have 2 independent mean angles instead of one “really good” mean angle that might’ve gotten screwed by instrument centering errors.
Every corner on the lots where shot at minimum twice from different set ups as we traversed .
+1. I preach the same: observing control points/monuments from 2 different setups is a requirement for me. One method I recommend is a resection using 2+ control points and whatever monuments you can see- resection geometry isn’t necessarily important… it’s observing additional relationships between your control and monuments.
I was taught years ago that that boundary survey measurements wise is only as good as your worst side shot.
Also +1. There are VERY FEW office techs that take a moment to review processing and think about the quality and pedigree of the data that they’re using to determine their boundary. Most simply have their blinders on and start fiddling with linework until it “clicks”. Given this, it’s imperative that the data they’re provided is the best you can get them without blowing your budget to kingdom come.
@steinhoff I have not broke set up I usually run main traverse by forced centering method engrained in me from geodetic traverse. I do sometimes teach closing the horizon especially when i have a corner that cannot be tied from two different locations. In that case bs the fs wrap angles again. bs the corner turn all angles again.
I agree though with today??s equipment the 4 rounds are nat as important. If equipment is in adjustment. We no longer have to rotate the bottom plates and such. I do like cross ties and yes resection??s. They are actually great when using rtk data and needing to tie that together and ir have a corner in a non rtk place . Simply set up robot or total station and resect from control etc.
In the software i can choose to use mean angles or all observations depending on what the data looks lime i make my decision at that time.
most of the time it is not hard to observe a corner from a couple set ups. Or more. I learned most of that out west. In east you can barely get some to even turn a direct and reverse round. Some do some do not. Some can careless about the measurement quality some care greatly. But our measurements are only a small portion of the survey. I just like doing the best i can with the things i know i can control and eliminate as an issue. If i have confidence in my measurements. I can get on with the rat killen and look at the boundary as it relates to the title deed etc. Or more research.
I have not broke set up I usually run main traverse by forced centering method engrained in me from geodetic traverse.
This is fine, but running forced centering and then trying to run a LSA is just… pointless in my opinion. You’re artificially going to have a clean adjustment.
In the software i can choose to use mean angles or all observations depending on what the data looks lime i make my decision at that time.
For this one… Let’s see if I get flamed. If you run “all observations” (in TBC, or utilizing direction sets in Star*Net) the program is assigning centering errors to every single observation, and can potentially give you a cleaner adjustment than you really should have. Utilizing mean angles instead is a more faithful representation of what happened in the field. Why should every component of your rounds be given individual centering errors? There’s just the one from your singular setup when you ran rounds.
EDIT: don’t get me wrong, I use individual observations at times to see if there’s specific observations in a mean angle that might be considered an outlier. But at the same time if you’re checking your mean angle reports and it’s kosher relative to the rest of the components of your mean angle… should you really remove it? You’re starting to get into the territory of cherry picking your data.
In east you can barely get some to even turn a direct and reverse round. Some do some do not.
+1. Full disclosure I was “trained” that a singular “direct” measurement is the only way to go. I.E. I wasn’t even told about rounds/sets. Not proud of that, but here we are- it’s a pretty widespread problem if you ask me.
But our measurements are only a small portion of the survey.
For this one I respectfully disagree (although I’m taking the statement at face value instead of reading between the lines). Your measurements are EVERYTHING. If you can’t accurately capture the relationship between control/monuments in the field and your project, you’ve got nothing. Might as well be one of those companies that draw some lines on a Google Earth image and call it a day.
Of course one’s experience, proper documentation, etc. all are vital components of a survey and the resulting deliverable, but you’re literally sunk without your measurements.
I somehow managed to run a static session on an out of level tripod last week. Not sure if I brain malfunctioned or one of the station fire fighters accidentally kicked a leg over when walking by. It would??ve shown up in TBC but knowing it was bad I didn??t import it. Had plenty of other sessions on that point. TBC flagged an RTK observed control point I had a feeling might be bad due to controller messages in the field so I disabled it. That??s the point of processing data??to find bad data and disable it.
My elementary surveying book talks about least squares. What books do you guys recommend that foccus on error propagation and least squares adjustment?
@steinhoff well my words are probably not well stated. What i mean is the measurements are not the total package of deciding a boundary. I if i want could run a very very accurate and very precise survey. Now if that survey was done in chains and i am following that surveyor i don??t consider myself a better measurement person. I have to look at all the evidence if I measure different than that surveyor did with a 20? transit and chain. Am i to move his monument to where he said it should go. I don??t believe so if all evidence points to said monument and I accept that was what he set or called for and it passes the test. Why would my highly accurate and precise measurements be on top. Maybe that helps explain. What i meant.
now on force centering yes in a conventional type set up no cross checks the centering error could be artificial but if you are using rtk and a rod for cross checks in addition to force centering and making measurements from different locations and cross ties I believe you are fine. Tbc doesn??t allow me to weight every individual type like that as well as starnet did from what I remember. It has been many many years since i used starnet. Starnet also allowed for taping if I remember correctly in the network adjustment. I wish tbc did. I was at a chapter meeting not long ago where the discussion in part was eliminate the steel tape calibration in the regs. I asked them not to. Because in cases where you have corners less than a hundred feet it is a great way to apply in a least squares as extra measurements directly to each other that might not otherwise get measurements on not to many people will set up on a corner that has another monument 30 feet a part to measure. Both are usually tied from a different location. I do this often have several that had monuments on a line that were mere feet a part. I physically measure between them. Even though they were rtk or robot or both located.
I’m woefully deficient in this area. I just found something you may want to start with.
A Correctly Weighted Least Squares Adjustment, Part 1
https://www.xyht.com/surveying/a-correctly-weighted-least-squares-adjustment-part-1/
MHThe old StarNet manual is a pretty good primer on LSA. If someone has the PDF for upload.
What i mean is the measurements are not the total package of deciding a boundary.
Of course. Like I said I was cherry picking a specific sentence and taking it literally to just point out a personal gripe of mine… wasn’t intending to point anything at you.
Now if that survey was done in chains and i am following that surveyor i don??t consider myself a better measurement person.
It’s funny because my old man has said similar things to me when we discuss work (him retired county surveyor, me corporate slave). He’ll say something along the lines of what you said, and my response would be along the lines of “I’m not trying to piss on a dead guy’s shoes, I’m just trying to collect and report accurate data so we can make the most educated decisions we can.” I don’t think either rationale is wrong per se.
now on force centering yes in a conventional type set up no cross checks the centering error could be artificial but if you are using rtk and a rod for cross checks in addition to force centering and making measurements from different locations and cross ties I believe you are fine.
Folks like @rover83 would be able to better explain this than me, but including any amount of forced centering in a LSA has the potential of screwing you up. “Normal traversing” + RTK (minimum of double occupying your points with GNSS! I’ll die on that hill…) to control points and monuments can give you fantastic results on their own.
Tbc doesn??t allow me to weight every individual type like that as well as starnet did from what I remember.
This is correct. Star*Net allows you to weight individual observations (which can be VERY risky if an office tech with insufficient experience is weighting individual observations to artificially make things work). Meanwhile TBC calculates standard errors for mean angles as uses those in a LSA. You CAN however weight individual reference factors in TBC (same comment about inexperienced office techs applies here, as well), which essentially boils down to weighting individual types of observations.
Starnet also allowed for taping if I remember correctly in the network adjustment.
Sure does. Just input a measured distance and any associated standard errors and you’re off to the races. It’s a shame that TBC can’t do the same.
I do this often have several that had monuments on a line that were mere feet a part. I physically measure between them. Even though they were rtk or robot or both located.
For sure, not a bad idea at all. Personally I just ask crews to make sure we have enough redundant robotic and/or GNSS data and call it a day. But hey if you want to tape between points for an additional check or use it as fodder in Star*Net? Power to you.
Starnet also allowed for taping if I remember correctly in the network adjustment.
That’s very interesting. Unfortunately, taping has gone the way of the dodo.
MH@steinhoff well no offense at all. None was taken.
force centering you have the option to set it to zero on centering error so I truly don??t know how that would mess up a least squares. I see why and what breaking set up can do I understand that concept but at the same time you can introduce error. The other is no tribrach total station prism will sit perfectly in any other tribrach so force centering is still applicable and helps but i allow people to do what they wants. So even force centering still has a error it just mitigates it. Geodetic and metrology measurers truly expert measurers have probably been using these techniques before land surveyors lol. And least squares. Surveying is a small segment in metrology and we are not even trying to achieve what can be accomplished. Its like running levels the most important piece of running levels is the rod person (s). We can catch and account for instrument error the readings can be made fool proof almost through three wires. What we cannot do is make sure the rod is on the same exact point it was on when we fore sighted it.
i have much respect for @rover for sure. I think he is way smarter than i am. I have ran lots and lots of traverse many many miles worth. I have done triangulation and with distance as well. I don??t always but when i can i will force center. It is still getting a new set up as you run cross checks so getting redundant set ups over the point is still in effect. I have ran traverse and done boundary surveys to prove you can and never ever set a point in the ground. That set up is only a representation of that point on the ground. It is only necessary for convenience to come back to and use. It is only as good as when we last measured it . We do it every day. NGS terms these as passive vs active marks. Now days.
but you have some great points for sure is a great place to be here for all to learn from each other etc.
The old StarNet manual is a pretty good primer on LSA. If someone has the PDF for upload.
For anyone interested: Star*Net v.6 Manual
Forced centering is a great technique if we’re not going to be occupying those points again, and we don’t need rigorous statistical analysis of those ground points. The endpoints of a link traverse might be all we care about, and that’s fine.
From a statistical standpoint, in least squares analysis, each setup and the measurements flowing out from it is considered independent, and the observations will be evaluated as such. But when forced centering is employed in a traditional traverse, with the exception of the very first setup, the foresight tripod is the only “new” and independent part of each setup.
In other words, each time the crew moves up to the next station, they are are not introducing the normal (random) centering, levelling and measure-up errors that would result from moving the entire tribrach + TS/prism between setups – or at the very least releasing the tribrach from the tripod, moving it off the mark, then recentering and relevelling. (One could also argue that best practices would be to change the height of the tripod, but in practice I have found simply re-measuring the height is sufficient, as long as one inputs exactly what they read each time, even if it is a few millimeters different than the previous. This is in fact what I would expect to see, because there is again inherent error in those values.)
As a result, with respect to the ground points, forced centering will influence not only distance, but also angular measurements, the latter especially as distances get shorter. Although it sounds counterintuitive, rather than “removing error”, by forcing the exact same errors in a majority of the observations, we are merely hiding actual error (good error I would call it?) that we would otherwise be able to appropriately model, account for and weight.
When the data are run through a least-squares analysis, the high correlation between setups will (in a typical traverse with no blunders) result in a very low reference factor (especially for distances), and the data will always look far better than they actually are, at least with respect to the ground points that are ostensibly being measured.
I have heard some argue for reducing the centering error to zero (or near zero) in order to “account” for this, but that still doesn’t change the fact that the LSA routine will see multiple setups align nearly perfectly, and that when we move that instrument up, it could very well be that it or the backsight (or the next foresight) are off-center.
But we’ll never know because the data will always look nearly perfect with forced centering.
Does all of this matter for the average boundary survey? That all depends, said the surveyor. At the very least it muddles our adjustment routine and makes it difficult to assess problems with the data.
Again, if we’re never coming back to occupy those traverse points and we just need to use the traverse as a connection between two primary points or as a platform for measuring other points of interest exactly once, that’s absolutely fine. @olemanriver mentioned measuring common points from different setups, and that will definitely help strengthen everything – with respect to the theoretical points we are set up over, which are not the points on the ground but our tripod heads/bottom of tribrach.
Now, if we forgo putting points in the ground and compute our measure ups from the tripod head/bottom of tribrach, keeping that centering error at zero, we will generally see far more realistic results. That’s not what usually happens, as a traverse usually employs points placed in the ground. But as a wise PLS and professor of mine once said, forced centering is “surveying along the tops of your tripod”, so if one were to treat it that way it does make sense.
All that being said, I have more of a problem with forced centering as it applies to control networks rather than boundary work, because tolerances are far tighter. I have seen it used and abused in high-precision control work more times than I can count.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman@rover83 yes you are correct in a conventional type traverse loop etc. forced centering can if not done proper show unrealistic results. And I understand your point. Do not argue with it at all. Now a lot boils down as well on the field practices and equipment care. I have ran traverse were not all points in the ground are semi permanent kinda like a turn point for level run. But when i am force centering we all know from many phd papers that we have about 2mm of centering error on a tribrach. That is depending on type and care. I usually have multiple setups in some fashion on a certain percentage of those points from cross ties and or double run type scenarios as we no longer have to use an Invar tape as was in the past. I have 3 different eyeballs to check the centering on a point. The first person sets up over the point at fs. Then i man sets gun in and should ck again then when the bs is set it has eye balls on it again. I have seen the discussion go both ways. And both ways have a purpose for sure. I am not sold that least squares should not be done with force centering. I have read way to many papers calling for it to be used dr g. And many others. Remember that centering error is tribrach only. Now no prism and prism adapter is made perfect and none will set in the same tribrach as say the gun or anything else exactly the same. But i agree that breaking set up gives the opportunity to have more measurements to hopefully get to the truth. Kinda like dropping a plumb bob several times and making a dot we never ever drop it in same exact location. But this same approach can introduce more error as well. As I traverse conventional around a property i a. Usually throwing points to one side or the other as well when feasible so they can be at minimum tied from two or more set ups. I also like especially when i am also doing topo or locations like to occupy some of those points and traverse through when possible. So no fly shots but at-least two points observation etc. build triangles if i can. Dr G. Even has been quoted in papers as he wrote that it is a good technique for accuracy. I would have to look up as several people reference his writings in their own work. But force centering not done correctly can cause a lot of harm.
you are correct that the ref factor will be much lower in a force centering situation 9 times out of 10 so the distance seem to be better than what they possibly are but this is not a negative always as the equipment have become so much better. Sighting as well. USACE Caltrans NGS and others recommend the force centering techniques so it can??t be all bad
now the other thing is say on a boundary the boundary corners can never ever be more accurate than the control network/ traverse unless it becomes apart of the control network itself . I do strive to incorporate it into my control when possible and cost effective as we always have to weigh and balance both I have been told by many never use a property corner in your traverse i say why not are we not striving to measure the monuments that we are holding to the best of our ability yes some say but you cause issues if you don??t just side shoot the property corner. Now I don??t understand that logic but i am not the smartest . But if i am set up and i can measure the property corner and i corporate that into my LSA with confidence why not. But i am one that wants anything i can to be accurate and precise looking at all the ways i can measure it . Great write up Rover83. Wish i had your writing ability for sure
@eagle1215 It is likely the elementary surveying book you have is by particular author(s) that also produce an adjustments text as well. “Adjustment Computations: Statistics and Least Squares in Surveying and GIS” by Wolf and Ghilani is one. I’ve got an older copy, so maybe someone else can say if it is still in print or has newer editions.
@jon-payne
In school we used the Elementary Surveying book by Ghilani. I bought the mentioned book about adjustments. It should be here tomorrow.
Log in to reply.